柔性神经树模型的结构优化算法的改进

The Improvement of Structure Optimization Algorithm Based on Flexible Neural Trees

利用柔性神经树模型的改进结构优化算法对流程工业生产过程的参数进行筛选,在精确度最高的前提下在最短的时间内找到影响生产过程的重要参数,为流程工业生产过程控制提供理论依据。在柔性神经树模型的学习过程中,本算法的进化代数不是一个固定值,而是以误差率来控制进化代数,试验证明这种算法使模型最优,效率和精确度最高。柔性神经树模型的结构和参数优化分别由概率增强式程序进化和模拟退火算法完成。本文以水泥生产中重要生产过程之一的分解炉的生产过程为研究对象。研究结果表明本文提出的改进方法是非常有效的。

In this paper,we introduce an improved structure optimized algorithm based on flexible neural tree model and using it filter important parameters of fluid industry＇s production process.On the premise of the highest accuracy in the shortest time find important parameters impacting the production process,and provide theoretical basis for process control of fluid industry production.In the process of structural optimization,the number of evolution generation in this algorithm is not fixed,but using error rate control evolution.The structure and parameters of flexible neural tree are optimized by probabilistic incremental program evolution and simulation annealing respectively.We take production process of decomposing furnace as an example,one of important cement ＇s processes.The results show that the proposed method is effective and practical.